Type: url
Original: https://agilejourney.uzabase.com/entry/2025/08/29/103000
Status: generated
License: all-rights-reserved
Summary
A long-form interview published on Uzabase’s Agile Journey media in August 2025, in which Takuto Wada (t_wada) and agile coach Tsutomu Yasui (yattom) discuss what coding agents change about automated testing and test-driven development. They argue that generative AI has lowered both the learning cost and the implementation cost of writing automated tests — being especially strong at characterization tests that capture how existing code behaves — while test design remains human work that AI handles well only when told which technique to apply. Their central claim is that TDD has become a guardrail against agent output degrading: it combines “it works” and “it stays maintainable” in a single workflow, and the phrase “test-driven development” is an efficient way to carry that whole context to a model. The conversation also covers Scrum roles under AI, documentation-code divergence, and how engineers might grow and remain employable.